The dinucleotide CG as a genomic signalling module.

The operon model proposed the existence of a category of proteins that control gene expression by interacting with specific DNA sequences. Since then, a large number of transcription factors recognizing a diversity of sequence motifs have been discovered. This article discusses an unusually short protein recognition sequence, 5'CG, which is read by multiple DNA binding proteins. CG exists in three distinct chemical states, two of which bind mutually exclusively to proteins that modulate chromatin structure. Non-methylated CG, which is highly concentrated at CpG island promoters, recruits enzymes that create the mark of promoter activity, trimethyl-lysine 4 of histone H3. Methylated CG, on the other hand, is a gene silencing mark and accordingly recruits enzymes that deacetylate histones. Thus, CG, despite its simplicity, has the properties of a genome-wide signalling module that adds a layer of positive or negative control over gene expression.